On prediction of bursty traffic in broadband satcom systems

For the purpose of traffic control in broadband satcom systems, it is required to predict the traffic descriptor for the next two round trip delay. The predicted traffic descriptor is used by earth-stations to shape their traffic in order to minimize traffic loss in the downlink queue and maximize the uplink utility. The paper investigates the application of neural networks or adaptive neuro fuzzy inference systems (ANFIS) in the prediction of bursty traffic and also application of neural networks in deriving the traffic descriptor. The result shows that the neural network approach is adequate for real time prediction of the traffic descriptor.